A statistical analysis of cancer genome variation

A statistical analysis of cancer genome variation

M.A. Newton , T.R. Yeager, and C.A. Reznikoff.

First issued November 1997. Revised April 1998.


We consider separating signal and noise in data measuring genetic abnormalities of cancer tumor cells. The linked hot-cold model is introduced and fit to data from a comparative genomic hybridization study of bladder cancer. We analyze these data and report details of the calculation summarized in Yeager et al. 1998. The model-generating framework developed in Newton et al. 1998 is used to motivate the linked hot-cold model.

Keywords Comparative genomic hybridization, deletions, instability-selection model, linked hot-cold model, Poisson process, statistical genetics, suppressor gene.

For ``Statistics in the Health Sciences: Genetics.'' Proceedings 1997 Summer Program at the Institute of Mathematics and its Application, Minneapolis, MN, Betz Halloran, Seymour Geisser, eds.

Postscript copy